1de Duve Institute, UCLouvain, Brussels, Belgium
2Institute for Biomedicine, Eurac Research, Italy
3Department of Anaesthesiology and Intensive Care, University Medicine Greifswald, Germany
*order of authors defined by sample with a random seed of 42
https://doi.org/10.5281/zenodo.3566699

RforMassSpectrometry

  • Initiative to create a flexible and scalable infrastructure for MS data.
  • See Laurent Gatto’s poster for details.

Mass spectrum

  • Spectrum:
    • 2 numeric: m/z and intensity values.
    • additional metadata information.
  • MSnbase: Spectrum object for a single spectrum.

Mass spectrum

… but usually we deal with many spectra …

Think bigger: Spectra

  • One object to represent data from one or many spectra.

Think flexible: MsBackend

  • Separate user functionality from data handling and storage.

Think flexible: MsBackend

  • Separate user functionality from data handling and storage.
  • Enables use of different backends (in-memory/on-disk, remote files, SQL-based, …).

Example: data import

  • Import data from an mzML file.
library(Spectra)
library(magrittr)

sps <- Spectra("data/20191107_Mix2_CE20.mzML", backend = MsBackendMzR())
sps
## MSn data (Spectra) with 1255 spectra in a MsBackendMzR backend:
##        msLevel           rtime scanIndex
##      <integer>       <numeric> <integer>
## 1            1   0.27700000002         1
## 2            1   0.58000000002         2
## ...        ...             ...       ...
## 1254         1 480.32500000002      1254
## 1255         1 480.68899999998      1255
##  ... 33 more variables/columns.
## 
## file(s):
## 20191107_Mix2_CE20.mzML
## Processing:
##

Example: data subsetting

  • Select all MS2 spectra for a [M+H]+ ion of Histidine.
mz_hist <- 156.07675 # Histidine
ms2_hist <- sps %>%
    filterMsLevel(2) %>%
    filterPrecursorMz(mz = mz_hist + ppm(c(-mz_hist, mz_hist), 20))
ms2_hist
## MSn data (Spectra) with 1 spectra in a MsBackendMzR backend:
##     msLevel     rtime scanIndex
##   <integer> <numeric> <integer>
## 1         2   184.593       489
##  ... 33 more variables/columns.
## 
## file(s):
## 20191107_Mix2_CE20.mzML
## Processing:
##  Filter: select MS level(s) 2 [Sat Dec  7 18:07:26 2019]
##  Filter: select spectra with a precursor m/z within [156.073628465, 156.079871535] [Sat Dec  7 18:07:26 2019]

Example: data processing

ms2_hist <- ms2_hist %>%
    pickPeaks() %>%
    removePeaks(threshold = 500) %>%
    clean(all = TRUE)
ms2_hist
## MSn data (Spectra) with 1 spectra in a MsBackendMzR backend:
##     msLevel     rtime scanIndex
##   <integer> <numeric> <integer>
## 1         2   184.593       489
##  ... 33 more variables/columns.
## 
## file(s):
## 20191107_Mix2_CE20.mzML
## Lazy evaluation queue: 3 processing step(s)
## Processing:
##  Filter: select MS level(s) 2 [Sat Dec  7 18:07:26 2019]
##  Filter: select spectra with a precursor m/z within [156.073628465, 156.079871535] [Sat Dec  7 18:07:26 2019]
##  Peak picking with MAD noise estimation, hws = 2, snr = 0 [Sat Dec  7 18:07:26 2019]
##  Signal <= 500 in MS level(s) 2 set to 0 [Sat Dec  7 18:07:26 2019]
##  Spectra of MS level(s) 2 cleaned. [Sat Dec  7 18:07:26 2019]
  • Data manipulations applied on-the-fly.

Example: use custom backend

  • Import reference spectra from HMDB.
library(MsBackendHmdb)
fls <- dir("data/hmdb_all_spectra", full.names = TRUE, pattern = "ms_ms_")
hmdb <- Spectra(fls, source = MsBackendHmdbXml(), nonStop = TRUE)
hmdb
## MSn data (Spectra) with 458963 spectra in a MsBackendDataFrame backend:
##          msLevel     rtime scanIndex
##        <integer> <numeric> <integer>
## 1              2        NA        NA
## 2              2        NA        NA
## ...          ...       ...       ...
## 458962         2        NA        NA
## 458963         2        NA        NA
##  ... 21 more variables/columns.
## Processing:
##  Switch backend from MsBackendHmdbXml to MsBackendDataFrame [Tue Dec  3 16:19:53 2019]

Example: compare spectra

  • Match spectrum against database.
res <- compareSpectra(ms2_hist, hmdb, ppm = 40)
hmdb$compound_id[res > 0.7]
## [1] "HMDB0000177" "HMDB0000177"

See Sebastian Gibb’s poster for details